Google DeepMind CEO Demis Hassabis recently narrowed his prediction for when Artificial General Intelligence (AGI) might arrive, shifting the timeline from 2030-2035 to a startling 2029-2030. This acceleration in just under a year has tech leaders talking, reigniting the debate over whether we are truly at the precipice of, or even already in, the AI singularity.
Hassabis, a Nobel laureate known for his measured statements, described our current position as standing in the βfoothills of the singularity.β Heβs not alone in this assessment. Figures like Elon Musk, Anthropic CEO Dario Amodei, Stripe's Patrick Collison, and OpenAI's Greg Brockman have all made similar, increasingly urgent statements about AI's rapid evolution.
Key Takeaways
- DeepMind CEO Demis Hassabis moved his AGI prediction to 2029-2030, a significant acceleration.
- AI agents are transforming from conversational tools into operational software, handling complex tasks across business workflows.
- Machines are solving long-standing mathematical and scientific problems, speeding up discovery.
- The infrastructure supporting AI is evolving at an unprecedented pace, from hardware to fabrication.
- While definitions vary, many tech leaders suggest we are in or rapidly approaching a 'singularity' where AI transforms society completely.
DeepMind's Hassabis Speeds Up AGI Forecast
For Hassabis, the singularity isn't a distant, abstract event. He sees it as the era we are in right now, a period of transformation that could be β10 times the industrial revolution at 10 times the speed.β This isn't casual hyperbole; it's a claim from someone intimately involved in building these systems.
Why this matters
When a leader of Hassabis's caliber, typically known for restraint, makes such a dramatic shift in his forecast, it signals a profound change in the underlying technology. His personal experience of AI coding agents collapsing timelines for tasks that once took months into mere hours underscores the tangible acceleration.
AI Agents Move From Chatbots to Core Operations
The shift isn't just in predictions; it's in the capabilities of AI itself. AI agents are evolving rapidly. In 2026, they are moving beyond simple chatbots to become operational software embedded in real business environments. These agents can plan, sequence, and execute complex workflows across multiple systems and tools.
AWS recently added new payment capabilities for autonomous agents, allowing AI systems to complete transactions and take more direct action within enterprise workflows.
This means AI isn't just assisting; it's actively performing tasks that previously required human intervention. SAP's sustainability AI agents, for example, are already delivering measurable results, including over 50% reduction in packaging compliance review hours and scenario simulation time cut from a day to 20 minutes. These aren't minor tweaks; they are order-of-magnitude improvements in productivity.
Machines Solving Century-Old Science Problems
Perhaps the most compelling evidence of acceleration comes from the scientific community. A new system called Axiomrover has quietly made mathematical breakthroughs, with eight papers appearing on arXiv since February, five already accepted by peer-reviewed journals. It has proven century-old problems, like showing that 100% of primes are partially regular under certain conditions.
The Chan Zuckerberg Biohub released a
